CS 526

Final project

Proposal presentation: March 2-4
Proposal writeup due: March 6
Midway progress presentaiton: March 30-April 1
Final presentation:April 27-29
Final paper due: May 4

In this project, you will move beyond replication and design a novel research study in data visualization / VR. You will formulate original research questions, design and conduct a human-subject experiment or develop a perceptual model, analyze the resulting data, and write a full research paper. Unlike Project 1, which focused on replicating and extending prior graphical perception studies, this project asks you to:

• Identify an open research question in data visualization / VR
• Ground your question in prior literature
• Design and conduct a human-subject study or build and evaluate a perceptual/computational model
• Analyze results using appropriate statistical methods
• Write a research paper in academic format

Overview

Your project must investigate a research question in data visualization or VR and include at least one of the following:

• A human-subject evaluation (e.g., perception, cognition, decision-making, interaction, trust, bias)
• A perceptual or computational model that predicts user performance and is validated against empirical data

You may build on ideas from Project 1, but the research question, design, and analysis must go substantially beyond simple replication.

---

Step 1: Identify your central research question

Based on your reading + search of prior literature, identify an evaluation gap where there is a lack of empirical testing for one or more visualization techniques. For example, perhaps there is a gap in understanding how node-link diagrams vs. adjacency matrices work, and how these two representations perform under certain tasks. So your research question could propose to compare these representations in ways the literature does not currently provide sufficient knowledge about.

Your question must be grounded and informed by prior literature. You should build on previous work and extend its results, rather than repeating it.

Additional examples (again, these are just examples and you are encouraged to propose your own ideas here):

• Does animation affect change detection accuracy?
• Do uncertainty visualizations improve probabilistic reasoning?
• Does the graphical perception of chart changes when they are shown in an AR/VR environment?
• Does interaction (e.g., brushing, tooltips) improve memory retention?
• How do log vs. linear scales affect magnitude comparison tasks?
• Can a perceptual model predict the effectiveness of visual encodings?

Step 2: Lit review

You must conduct a structured literature review that includes:

• At least 10 peer-reviewed research papers
• Foundational perception work (e.g., graphical perception, visual encoding, attention, memory)
• At least 3 papers directly related to your chosen topic

Your review should summarize key findings, identify limitations or gaps, and explain how your work extends or challenges prior research.

Step 3: Experimental or model design

Choose from the following options:

Option A: Human-subject evaluation

Your study must include:

• Clearly defined task(s)
• Controlled manipulation of independent variable(s)
• Randomization or counterbalancing
• At least 15 participants (unless otherwise approved)
• Multiple trials per condition

You must measure at least one quantitative dependent variable, such as: Accuracy, Response time, Bias, Confidence, Decision quality.

Option B: Perceptual or computational model

If you choose a modeling approach, you must:

• Formally define the model
• Explain its theoretical basis
• Generate quantitative predictions
• Validate predictions against empirical data (collected by you or from prior studies)

Step 4: Implementation, data collection and analysis, and results presentation

You may build experimental interfaces or models using D3, Observable, Python, or another appropriate platform.

Your analysis must go beyond descriptive statistics. Depending on your design, this may include: confidence intervals, hypothesis testing (e.g., t-tests, ANOVA, regression), and model fitting and goodness-of-fit evaluation. In the presentation of results, you must clearly represent uncertainty (e.g., show confidence intervals in all figures).

Step 5: Write a final research paper

Your final deliverable is a research paper written in academic format (approximately 6–10 pages, single-spaced equivalent).

The paper must include the following sections:

• Abstract: The abstract should concisely summarize your research question, methods, results, and contributions.
• Introduction: Motivate the research problem and questions. Summarize your contributions.
• Related work: Literature review and positioning of your work.
• Research questions and hypotheses: Clearly stated research questions and formal hypotheses. You should also explain where your hypotheses and predictions come from, rather than just stating them.
• Methods: Experimental design, participants, apparatus, stimuli, procedure, measures, and analysis plan.
• Results: Statistical analysis, figures, effect sizes, and interpretation.
• Discussion: Interpretation of findings, limitations, implications for visualization design.
• Conclusion / future work

Your paper should read like a submission to a venue such as IEEE VIS or CHI.

Deliverables

• You should first propose your project by presenting to the class during the week of March 2-4. Your presentation must include the research question and significance of the proposed work, a summary of how prior literature approached this question, and a description of the methods you plan to use.
• Based on feedback, you should send me a proposal (2-3 pages). The proposal should include the above elements. That is: research questions, significance, brief lit review, and proposed methods
• Mid-project presentation where you will present preliminary results of the project to the class
• Final project presentation during the week of April 27-29 (25 minutes each). At this point, you should have most of your results collected and analyzed.
• Final paper (6–10 pages) due on May 4th.

Grading criteria

Grading will be based on:

• Quality of your proposal
• Clarity and significance of research question
• Grounding in prior literature
• Soundness of experimental or modeling methodology
• Strength of discussion and insight
• Quality of your writing and presentation